Conserved Bacterial Genomes from Two Geographically Distinct Peritidal Stromatolite Formations Shed Light on Potential Functiona

1 Conserved bacterial genomes from two geographically distinct peritidal

2 stromatolite formations shed light on potential functional guilds

4 Running Title: Functional bacterial guilds in stromatolites

6 Samantha C. Waterwortha, Eric W. Isemongerb, Evan R. Reesa, Rosemary A.

7 Dorringtonb and Jason C. Kwana#

8 a Division of Pharmaceutical Sciences, University of Wisconsin, 777 Highland Ave.,

9 Madison, Wisconsin 53705, USA

10 b Department of Biochemistry and Microbiology, Rhodes University, Grahamstown,

11 South Africa

12 #Corresponding author. Email: [email protected]; Phone: 608-262-3829

14 ORIGINALITY-SIGNIFICANCE

15 Peritidal stromatolites are unique among stromatolite formations as they grow at the

16 dynamic interface of calcium carbonate-rich groundwater and coastal marine waters.

17 The peritidal space forms a relatively unstable environment and the factors that

18 influence the growth of these peritidal structures is not well understood. To our

19 knowledge, this is the first comparative study that assesses species conservation within

20 the microbial communities of two geographically distinct peritidal stromatolite

21 formations. We assessed the potential functional roles of these communities using

22 genomic bins clustered from metagenomic sequencing data. We identified several

23 conserved bacterial species across the two sites and hypothesize that their genetic

24 functional potential may be important in the formation of pertidal stromatolites. We

25 contrasted these findings against a well-studied site in Shark Bay, Australia and show

26 that, unlike these hypersaline formations, archaea do not play a major role in peritidal

27 stromatolite formation. Furthermore, bacterial nitrogen and phosphate metabolisms of

28 conserved species may be driving factors behind lithification in peritidal stromatolites.

30 SUMMARY

31 Stromatolites are complex microbial mats that form lithified layers and ancient forms are

32 the oldest evidence of life on earth, dating back over 3.4 billion years. Modern

33 stromatolites are relatively rare but may provide clues about the function and evolution

34 of their ancient counterparts. In this study, we focus on peritidal stromatolites occurring

35 at Cape Recife and Schoenmakerskop on the southeastern South African coastline.

36 Using assembled shotgun metagenomic data we obtained 183 genomic bins, of which

37 the most dominant taxa were from the Cyanobacteriia class (Cyanobacteria phylum),

38 with lower but notable abundances of bacteria classified as Alphaproteobacteria,

39 Gammaproteobacteria and Bacteroidia. We identified functional gene sets in bacterial

40 species conserved across two geographically distinct stromatolite formations, which

41 may promote carbonate precipitation through the reduction of nitrogenous compounds

42 and possible production of calcium ions. We propose that an abundance of extracellular

43 alkaline phosphatases may lead to the formation of phosphatic deposits within these

44 stromatolites. We conclude that the cumulative effect of several conserved bacterial

45 species drives accretion in these two stromatolite formations.

47 INTRODUCTION

48 Stromatolites are organo-sedimentary structures that date back more than 3.4 billion

49 years, forming the oldest fossils of living organisms on Earth (Dupraz et al., 2009). A

50 more recent discovery of stromatolite-like structures in Greenland suggests that the

51 structures may date as far back as 3.7 - 3.8 billion years (Nutman et al., 2016),

52 however, their biogenic origin remains under debate (Witze, 2016). The emergence of

53 Cyanobacteria in stromatolites approximately 2.3 billion years ago initiated the Great

54 Oxygenation Event that fundamentally altered the Earth’s redox potential and resulted in

55 an explosion of oxygen-based and multicellular biological diversity (Soo et al., 2017).

56 Ancient stromatolites could provide insight into how microorganisms shaped early

57 eukaryotic evolution. Unfortunately ancient microbial mats are not sufficiently preserved

58 for identification of these microbes and individual bacteria cannot be classified more

59 specifically than phylum Cyanobacteria due to morphological conservatism (Awramik,

60 1992; Dupraz et al., 2009). The study of extant stromatolite analogs may therefore help

61 to elucidate the biological mechanisms that led to the formation and evolution of their

62 ancient ancestors. Modern stromatolites are formed through a complex combination of

63 both biotic and abiotic processes. The core process revolves around the carbon cycle

64 where bacteria transform inorganic carbon into bioavailable organic carbon for

65 respiration. Bacterial respiration in turn results in the release of inorganic carbon, which,

66 under alkaline conditions, will bind cations and precipitate primarily as calcium

67 carbonate (Dupraz et al., 2009). This carbonate precipitate, along with sediment grains,

68 can then become trapped within bacterial biofilms forming the characteristic lithified

69 layers.

71 Alteration of the pH and subsequently, the solubility index (SI), may promote

72 mineralization or dissolution of carbonate minerals through microbial cycling of redox

73 sensitive compounds such as phosphate, nitrogen, sulfur and other nutrients within the

74 biofilm. This in turn regulates the rate of carbonate accretion and stromatolite growth.

75 Particularly, photosynthesis and sulfate reduction have been demonstrated to increase

76 alkalinity thereby promoting carbonate accretion, resulting in the gradual formation of

77 lithified mineral layers (Dupraz et al., 2009). In some stromatolite formations such as

78 those of Shark Bay, Australia, there is abundant genetic potential for both dissimilatory

79 oxidation of sulfur (which may promote dissolution under oxic conditions and

80 precipitation under anoxic conditions) and dissimilatory reduction of sulfate (which

81 promotes precipitation) (Gallagher et al., 2012; Casaburi et al., 2016; Wong et al.,

82 2018).

84 The biogenicity of stromatolites has been studied extensively in the hypersaline and

85 marine formations of Shark Bay, Australia and Exuma Cay, Bahamas, respectively

86 (Khodadad and Foster, 2012; Babilonia et al., 2018). The presence of Archaea has

87 been noted in several microbial mat and stromatolite systems (Casaburi et al., 2016;

88 Balci et al., 2018; Medina-Chávez et al., 2019), particularly in the stromatolites of Shark

89 Bay, where they are hypothesized to potentially fulfill the role of nitrifiers and

90 hydrogenotrophic methanogens (Wong et al., 2017). Although Cyanobacteria,

91 Proteobacteria and Bacteroidetes appear to be abundant in both marine and

92 hypersaline systems, Cyanobacteria are proposed to be particularly vital to these

93 formations through the combined effect of biofilm formation, carbon fixation, nitrogen

94 fixation and endolithic (boring) activity (Macintyre et al., 2000; Khodadad and Foster,

95 2012; Casaburi et al., 2016; Babilonia et al., 2018).

97 Peritidal tufa stromatolite systems are found along the southeastern coastline of South

98 Africa (SA). They are geographically isolated, occurring at coastal dune seeps

99 separated by stretches of coastline (Smith et al., 2018). In these systems, stromatolite

100 formations extend from freshwater to intertidal zones and are dominated by

101 Cyanobacteria, Bacteroidetes and Proteobacteria (Perissinotto et al., 2014). The

102 stromatolites are impacted by fluctuating environmental pressures caused by periodic

103 inundation by seawater, which affects the nutrient concentrations, temperature and

104 chemistry of the system (Rishworth et al., 2016). These formations are characterized by

105 their proximity to the ocean, where stromatolites in the upper formations receive

106 freshwater from the inflow seeps, middle formation stromatolites withstand a mix of

107 freshwater seepage and marine over-topping, and lower formations are in closest

108 contact with the ocean (Perissinotto et al., 2014). The stromatolite formations at Cape

109 Recife and Schoenmakerskop are exposed to both fresh and marine water that has little

110 dissolved inorganic phosphate and decreasing levels of dissolved inorganic nitrogen

111 (Cape Recife: 82 - 9 µM, Schoenmakerskop: 424 - 14 µM) moving from freshwater to

112 marine influenced formations (Rishworth, Perissinotto, Bird, et al., 2017). Microbial

113 communities within these levels therefore likely experience distinct environmental

114 pressures, including fluctuations in salinity and dissolved oxygen (Rishworth et al.,

115 2016). While carbon predominantly enters these systems through cyanobacterial carbon

116 fixation, it is unclear how other members of the stromatolite-associated bacterial

117 consortia influence mineral stratification resulting from the cycling of essential nutrients

118 such as nitrogen, phosphorus and sulfur. Since peritidal stromatolites exist in constant

119 nutritional and chemical flux with varying influence from the fresh and marine water

120 sources, they present an almost ideal in situ testing ground for investigating which

121 microbes are consistently present despite fluctuations in their environment. Identification

122 of conserved bacterial species across both time and space and across varied

123 environmental pressures would suggest that these bacteria are not only robust but likely

124 play important roles within the peritidal stromatolite consortia.

125

126 Previous studies of stromatolites collected from Lake Clifton, Pavilion Lake, Clinton

127 Creek, Highborne Cay and Pozas Azules have investigated the overall bacterial

128 composition using 16S rRNA gene analysis and/or overall potential gene function of

129 unbinned metagenomic datasets per bacterial taxa using tools such as MG-RAST

130 (Keegan et al., 2016) to gain insight into the potential roles of the different bacterial

131 groups (Mobberley et al., 2015; Centeno et al., 2016; Gleeson et al., 2016; Ruvindy et

132 al., 2016; Warden et al., 2016; White et al., 2016). However, the presence of all genes

133 required for a complete functional pathway within a collection of bacteria does not

134 necessarily mean that the cycle can take place since they may not be present within a

135 single organism. Therefore, in order to assess the potential for functional roles,

136 individual bacterial genomes must be investigated. To date, only two studies have

137 successfully binned individual bacterial genomes from culture-independent,

138 metagenomic data originating from stromatolites from Shark Bay (Australia) (Wong et

139 al., 2018) and Socompa Lake (Argentina) (Kurth et al., 2017). The latter study obtained

140 four high-quality bins (in accordance with MIMAG standards defined by (Bowers et al.,

141 2017)) and analyzed three which were believed to have carried genes from several

142 closely-related genomes. Extracted 16S rRNA sequences were in conflict with whole-

143 genome taxonomic classifications (Kurth et al., 2017). The Shark Bay study obtained a

144 total of 550 binned genomes, of which 87 (15.8%) were of medium to high quality

145 (Bowers et al., 2017; Wong et al., 2018) and the data provided information on a number

146 of potentially important processes that may contribute to the formation and maintenance

147 of the hypersaline stromatolites. However, the study did not identify key microbial

148 architects within these biogenic structures.

149

150 Using a metagenomic approach, we have sought to gain insight into the foundational

151 bacteria responsible for metabolic processes that potentially result in formation of

152 pertidal South African stromatolites. We obtained and annotated 183 putative bacterial

153 metagenome-assembled genomes (MAGs), (of which 112 (61%) were of medium to

154 high quality) from samples of two geographically isolated sites near Port Elizabeth,

155 South Africa. We identified several temporally and spatially conserved bacterial species

156 and functional gene sets, that are likely central in establishing and maintaining peritidal

157 stromatolite microbial communities.

158

159 RESULTS AND DISCUSSION

160 Two geographically isolated peritidal stromatolite sites, Cape Recife and

161 Schoenmakerskop, which are 2.82 km apart, were chosen for this study. These two

162 sites have been extensively characterized with respect to their physical structure,

163 nutrient and chemical environment (Perissinotto et al., 2014; Rishworth et al., 2016,

164 2019; Rishworth, Perissinotto, Bornman, et al., 2017; Dodd et al., 2018). The sites

165 experience regular shifts in salinity due to tidal overtopping and groundwater seepage

166 (Rishworth et al., 2019). Comparison to a site with groundwater seepage but no

167 stromatolite growth, has shown that the growth of peritidal stromatolites is promoted

168 within this region by decreased levels of wave action, higher water alkalinity and

169 decreased calcite and aragonite saturation (Dodd et al., 2018). Furthermore,

170 stromatolite growth is inhibited by increased levels of salinity, as lithified structures are

171 not observed in both the marine waters and in deeper portions of formation pools, with

172 higher salinity levels (Dodd et al., 2018).

173

174 Stromatolite formations at both sites begin at a freshwater inflow and end before the

175 subtidal zone (Fig. 1) and are exposed to different levels of tidal disturbance. Samples

176 for this study were collected from the upper stromatolites at Cape Recife and

177 Schoenmakerskop in January and April 2018 for comparisons over time and geographic

178 space. Additional samples were collected in April 2018 from middle and lower

179 formations for extended comparison across the two sites (Fig. 1). For a detailed

180 perspective of depth and the differentiation of the sampled zones a 3-dimensional

181 rendering of the sample sites was constructed by a 3rd party (Caelum Technologies)

182 using a combination of drone-based photogrammetry and geographic mapping using

183 differential GPS (Fig. 1). Stromatolite formations were classified according to their tidal

184 proximity as defined in previous studies (Perissinotto et al., 2014; Rishworth et al.,

185 2016; Rishworth, Perissinotto, Bornman, et al., 2017; Dodd et al., 2018). Upper

186 formations occur in the supratidal zone where they are constantly exposed to fresh

187 water flowing from dune seeps. These formations will only be exposed to seawater

188 during spring tides or extreme storm surges. The water from upper formations feeds into

189 large pools, which in turn feed into the lower portion of the system. In the upper-middle

190 intertidal zone, the formations form a slope into large pools where they will only receive

191 seawater during peak high tide. Lower formations occur in areas where saline or

192 brackish conditions are predominant. In Schoenmakerskop the lower zone is located in

193 a semi stagnant pool which is frequently overtopped, whilst in the lower zone of Cape

194 Recife, formations create a low flowing slope which ends at the subtidal zone. All

195 sampling was conducted at low tide. Sample abbreviations and MAG prefixes used

196 throughout this study correspond to the site and region from which they were sampled

197 e.g. “SU” identifies the sample as originating from Schoenmakerskop, Upper formation

198 in April (Table S1). Samples prefixed with a “C” identify samples collected in January

199 e.g. “CSU” identifies the sample as originating from Schoenmakerskop, Upper formation

200 in January (Table S1).

201

202 Phylogenetic distribution of microbial communities

203 We assessed the diversity and structure of the bacterial communities in triplicate

204 samples taken from upper, middle and lower formations at Cape Recife and

205 Schoenmakerskop using 16S rRNA gene amplicon sequence analysis. All communities

206 were dominated by Cyanobacteria, Bacteroidetes, Alphaproteobacteria,

207 Gammaproteobacteria and other unclassified bacteria (Fig. 2A), in agreement with a

208 previous study at Schoenmakerskop (Perissinotto et al., 2014). Bacterial communities at

209 each sampling site were statistically different from one another (Fig. 2B; ANOSIM: R =

210 0.976, p = 0.01 permutations = 999), whilst pairwise Kendall rank correlation tests

211 showed that replicates were not statistically different from one another (pairwise p-

212 values < 2.2 × 10-16).

213

214 Trends observed in metagenomes of stromatolite samples from Cape Recife and

215 Schoenmakerskop

216 To characterize the metabolic potential within stromatolites, we generated shotgun

217 metagenomic libraries from 8 samples representative of the upper, middle and lower

218 formations at both sites in April 2018, as well as an additional two samples from the

219 upper formations of each site, collected 4 months previously in January 2018 (Table

220 S1). Following assembly of raw sequence reads, gene coverages were normalized to

221 the length-weighted average coverage per sample, revealing a high abundance of

222 genes encoding phosphate transport (pstSCAB), phosphate uptake regulation

223 (phoURBP) and alkaline phosphatases (phoADX) were observed across the board (Fig.

224 3). Additionally, genes involved in phosphonate metabolism (phnCDEFGHIJKLM) were

225 abundant in three of the four middle formations but were absent in upper formations.

226 The concentration of soluble phosphorus, although relatively low, is highest in water

227 surrounding the middle formations, as both fresh seep water and ocean overtopping

228 contribute to the total phosphate (Dodd et al., 2018). It is thus unsurprising then, that the

229 greatest abundance of phosphate-metabolism genes are found in the middle

230 formations. The high abundance of genes encoding phosphate-metabolizing enzymes,

231 may also be indicative of how stromatolite communities cope with low dissolved

232 inorganic phosphorus in their environment. Genes encoding alkaline phosphatase

233 phoX, were the most abundant among the phosphatases observed in these stromatolite

234 samples, and represent a calcium-dependent enzyme that can function at low substrate

235 concentrations on a broad range of C-O-P substrates (Zaheer et al., 2009). The

236 phosphonate transporter genes (phnCDE) are more prevalent than the genes encoding

237 the C-P lyase (phnGHIJLM) required for phosphonate degradations (Fig. 3). These

238 transporters have also been implicated in the transport of inorganic phosphate, in

239 addition to phosphonates (Stasi et al., 2019), and the discrepancy between transporters

240 and metabolic genes may suggest an additional role of phosphate uptake in these

241 systems. Overall gene abundances indicated negligible presence of canonical

242 dissimilatory sulfate reduction/oxidation via aprAB and dsrAB encoded enzymes.

243 Similarly, there were low abundances of genes associated with sulfonate metabolism.

244 There was an abundance of genes associated with assimilatory sulfate reduction, but

245 the low abundance or absence of cysC, a gene encoding a key enzyme in this pathway,

246 indicated that this pathway may not be complete. Genes associated with assimilatory

247 nitrate reduction (narB, nirA) were the most prevalent markers of nitrogen metabolism.

248 All sites also contained a number of genes associated with dissimilatory nitrate

249 reduction where cytoplasmic NADH-dependent nitrate reductase nirBD appeared to be

250 favored over periplasmic cytochrome c nitrate reductase nrfAH. Interestingly nitrogen

251 fixation genes (nifDHK) appear to be more abundant in the middle stromatolite

252 formations of Schoenmakerskop (Fig. 3).

253

254 Binning and phylogenetic classification of putative genomes

255 The 10 assembled metagenomes were binned using Autometa (Miller et al., 2019),

256 resulting in a total of 183 bacterial genome bins (Table S1). Using relative coverage per

257 sample as a proxy for abundance, we found that genomes classified within the

258 Cyanobacteriia class were consistently dominant in all collection points, while

259 Alphaproteobacteria, Gammaproteobacteria and Bacteroidia were less abundant but

260 notable bacterial classes (Fig. 4A). This distribution appears to be approximately

261 congruent with abundances observed in the 16S rRNA gene amplicon analyses (Fig. 2).

262

263 Temporal and spatial conservation of bacterial species

264 We calculated pairwise average nucleotide identity (ANI) between all binned genomes

265 and defined conserved species as genomes sharing more than 97% ANI in two or more

266 of the sampled regions. We identified 16 conserved taxa across the 10 sampled

267 regions, with several species identified in 3 - 5 samples, across Schoenmakerskop and

268 Cape Recife, however no single species was common to all sampled sites (Table S2,

269 Fig. 4A). Conserved species were commonly the most abundant taxa present in each of

270 the samples, accounting for approximately 30 - 80% of the species abundance (Table

271 S2, Fig. 4A). Cyanobacterial species within the Acaryochloris and Hydrococcus genera

272 and Absconditabacterales family were conserved across upper formations (Table 1, Fig.

273 4B, Table S3). Seven other species were conserved across the middle formations,

274 including species classified within the Rivularia genus and Phormidesmiaceae and

275 Spirulinaceae families (Table 1 and S3, Fig. 4B). Two distinct species within family

276 Elainellaceaee were also conserved: species A was detected only in the upper pools of

277 Cape Recife, whilst species B was conserved across the middle formations of both sites

278 (Table 1 and S3, Fig. 4B).

279

280 Also, of interest was the presence of conserved bacterial species (order

281 Absconditabacterales), which are classified under the Patescibacteria phylum.

282 Patescibacteria are unusually small bacteria found in groundwater that produce large

283 surface proteins hypothesized to help them attach to, and exploit the ability of other

284 microorganisms performing nitrogen, sulfur and iron cycling (Herrmann et al., 2019).

285 The presence of these conserved bacteria suggests that the inflow water seeps may

286 originate from groundwater.

287

288 Conserved Rivularia, Elainellaceae, Phormidesmiaceae and Chloroflexaceae species

289 were identified in the lower formation of Schoenmakerskop, but none of the genomes

290 identified in the lower formation of Cape Recife had greater than 97% shared ANI with

291 any other genome bin. The distribution of conserved taxa, wherein the upper and middle

292 formations appear to harbor distinct conserved taxa suggests that the differing nutrient

293 and physical characteristics between upper, middle and lower regions of the

294 stromatolite formation elicit specialization of the conserved bacterial community. The

295 lack of conserved bacterial species across the lower formation of Cape Recife may be

296 due to the choice of sampling site: The lower formation sample taken from Cape Recife

297 is in closer contact with the ocean than the lower formation sample taken from

298 Schoenmakerskop (Fig. 1 C - D). The proximity of the Schoenmakerskop lower

299 formation sample to the ocean was initially thought to be sufficient to delimit it as a

300 lower formation, but following this study it may be reclassified as a middle formation, as

301 the bacterial community composition corresponds more closely with the other four

302 middle formation samples (Fig. 4B).

303

304 Metabolic potential of binned genomes

305 Oxygenic photosynthesis by cyanobacteria results in rapid fixation of carbon dioxide

306 and an increase in alkalinity (Pace et al., 2018). Carbonate ions bind cations such as

307 calcium and are precipitated under alkaline conditions, promoting the growth of

308 stromatolite structures (Dupraz et al., 2009). Given their numerical dominance in Cape

309 Recife and Schoenmakerskop stromatolites, and their predicted role in other

310 stromatolites (Dupraz et al., 2009), cyanobacteria likely perform carbon sequestration.

311 However, the identity of the bacteria that cycle redox-sensitive sulfur, phosphate,

312 nitrogen and calcium, and subsequently affect the alkalinity and solubility index enabling

313 carbonate precipitation in these stromatolites remains unknown. We inspected

314 PROKKA and KEGG annotations within all stromatolite-associated bacterial genome

315 bins to identify potential metabolic pathways that may promote mineral deposition and

316 accretion (Dupraz et al., 2009). The results presented here are summarized in Table 1.

317

318 Sulfur metabolism. Reduction of sulfate has previously been shown to promote the

319 precipitation of carbonates in the form of micritic crusts in Bahamian and Australian

320 stromatolites (Reid et al., 2000; Wong et al., 2018) and it has been suggested that

321 microbial cycling of sulfur played an important role in ancient Australian stromatolites,

322 even prior to the emergence of Cyanobacteria (Bontognali et al., 2012; Allen, 2016).

323 Amongst the Cape Recife and Schoenmakerskop stromatolite-associated bacteria, the

324 potential capacity for sulfate reduction was confined to only a few genomes (Fig. 5 and

325 Fig. S1). The complete set of genes required for assimilatory sulfate reduction

326 (sat/met3, cysC, cysH and sir genes) (Santos et al., 2015) were recovered in four

327 genomes (Fig. S1), three of which were conserved Acaryochloris species (Fig. 5). The

328 abundance of genes associated with assimilatory sulfate reduction in the conserved

329 Acaryochloris genome bins accounted for 22 - 100% of gene counts observed in their

330 respective metagenomes (Fig.5). This was calculated as abundance of geneX in binX,

331 as a percentage of geneX abundance in the sample from which the bin was derived

332 (Fig. 5). The greatest abundance genes for uptake and desulfonation of

333 alkanesulfonates (ssuABCDE) (Aguilar-Barajas et al., 2011; Ellis, 2011) were detected

334 exclusively in conserved Hydrococcus species (Fig. 5), which account for 23 - 100% of

335 gene abundance in the respective metagenomes. All genes required for a complete

336 pathway found exclusively in bin RU2_2, but both SU_1_0 and CRU1_1 appear to be

337 missing the ssuE gene. The ssuE gene is not required for growth using aliphatic

338 sulfonate or methionine substrates, but is required for arylsulfonate metabolism

339 (Kahnert et al., 2000). This suggests that these Hydrococcus strains are all capable of

340 some form of sulfonate metabolism, but not all can metabolize arylsulfonates (Bin

341 CSU_1_8 is only 37% complete, and therefore of low quality and may be missing gene

342 due to incompleteness). In both cases, these trends are in agreement with the patterns

343 observed in the metabolic potential of the overall metagenome (Fig.3). Hydrococcus

344 and Acaryochloris species are dominant in upper formations (Fig. 4 and Table S3) and

345 the potential for cumulative removal of hydrogen by sulfate reduction by these species

346 may aid in the creation of an alkaline environment within the system. Seep waters

347 feeding both Cape Recife and Schoenmakerskop have relatively high levels of sulfate

348 (Dodd et al., 2018), and it is possible that the Hydrococcus and Acaryochloris species

349 utilize this nutrient in the upper formations (closest to the seeps) as a selective

350 advantage, simultaneously increasing the alkalinity of the surrounding environment

351 through hydrogen assimilation. There was no evidence for the potential for dissimilatory

352 sulfate reduction in conserved genome bins (Fig. 5). Similarly, amongst all 183 genome

353 bins, only genomes representative of a Thioiploca sp. from the Schoenmakerskop lower

354 formation and a Desulfobacula sp. from the Schoenmakerskop middle formation

355 included all genes required for dissimilatory sulfate reduction (Fig. S1). The lack of

356 genes associated with conserved or dominant bacterial taxa in the middle and lower

357 pools suggest that either sulfate is not required by the consortia that inhabit the

358 middle/lower formations and is required only by bacteria exposed to ground water, or

359 that insufficient amounts of the sulfate continues into the lower pools where the middle

360 and lower formations are found. This suggests that calcite formation in these peritidal

361 stromatolites may be influenced by processes other than dissimilatory sulfate reduction,

362 in contrast to stromatolite formations in the Cayo Coco lagoonal network, Highborne

363 Cay and Eleuthera Island in the Bahamas (Visscher et al., 2000; Dupraz et al., 2004;

364 Pace et al., 2018).

365

366 Nitrogen metabolism. The reduction of nitrogen, nitrates and nitrites can lead to calcite

367 precipitation (Rodriguez-Navarro et al., 2003; Wei et al., 2015; Konopacka-Łyskawa et

368 al., 2017; Wong et al., 2018; Lee and Park, 2019), and the released NH3 can react with

4+ 2- 369 CO2 and H2O, to form 2NH + CO3 (Konopacka-Łyskawa et al., 2017). Nitrogen

370 metabolism is proposed to have emerged at approximately the same time as sulfur

371 metabolism, dated to 3.4 billion years ago (Stüeken, 2016), as both processes share

372 similar redox states (Thomazo et al., 2011), and ammonium availability during this time

373 may have sustained developing microbial life (Yang et al., 2019). Therefore, bacteria

374 that can fix nitrogen or produce ammonia from nitrates/nitrites could potentially promote

375 the growth of stromatolites (Visscher and Stolz, 2005) and may have added to the

376 formation of ancient analogs. Similarly, denitrification will likely lead to mineral

377 dissolution and retardation of stromatolite growth (Visscher and Stolz, 2005). We found

378 that several conserved bacteria carried the genes associated with ferredoxin-dependent

379 assimilatory nitrate reduction (nirA-narB genes) (Moreno-Vivián et al., 1999), nitrogen

380 fixation (nifDHK genes) and dissimilatory nitrite reduction (nirBD genes) (Griffith, 2016),

381 all of which result in the production of ammonia (Fig. 5).

382

383 The potential for assimilatory nitrate reduction was detected in several genomes,

384 primarily from the middle formations, with particularly high gene abundances of nirA-

385 narB genes conserved Rivularia sp. (Fig. 5), which are the dominant species in the

386 middle formations (Fig. 4). The potential for dissimilatory reduction of nitrites, via either

387 cytoplasmic nirBD or membrane-bound nrfAH nitrite reductases, was detected in

388 several genomes across both the upper and middle formations (Fig. S1). It appears that

389 the gene abundances associated with assimilatory nitrate reduction in Rivularia sp.

390 account for the majority (29 - 75%, Fig. 5) of nirA-narB genes observed in their

391 respective metagenomes (Fig. 3). Among conserved bacteria, the potential for

392 dissimilatory nitrate reduction is relatively low, with Hydrococcus sp. carrying the

393 greatest abundance of nirBD genes (Fig. 5) which encode cytoplasmic nitrite

394 reductases. Non-conserved Thioploca, Cyclobacteriaceae and Limnothrix species

395 appear to account for the remainder of the nirBD genes observed in their respective

396 metagenomes (Fig. S1). Very few genomes included the nrfAH genes which encode

397 membrane-bound nitrite reductases.

398

399 Nighttime nitrogen fixation has previously been shown to be a driver of carbonate

400 precipitation in stromatolites (Dupraz et al., 2009) and nif genes were identified in

401 several conserved bacteria associated with both upper and middle formations:

402 Hydrococcus species across both upper formations, Microcoleus in the

403 Schoenmakerskop upper formations and Phormidiaceae, Spirulinaceae and

404 Chloroflexaceae species the middle formations (Fig. 5). In addition to the conserved

405 bacteria carrying nif genes, non-conserved Blastochloris species appeared to account

406 for the majority (10% - 75%) of identified nifDHK genes (Fig. S1). The genetic capacity

407 for nitrogen fixation being retained in Schoenmakerskop and Cape Recife was

408 unexpected, given the high concentrations of dissolved inorganic nitrogen (DIN) ranging

409 from 95 - 450 mM at the two sites (Rishworth et al., 2016).

410

411 Phosphate and phosphonate metabolism

412 Phosphatic structures that closely resemble fossilized phosphatic stromatolites have

413 recently been observed within Cape Recife stromatolites (Buttner et al., 2019,

414 unpublished). Hydroxyapatite (Ca5(PO4)3(OH)) is more easily precipitated than calcite

3- 415 (CaCO3) and the release of PO4 into the biofilm by alkaline phosphatase activity could

416 increase the potential for apatite mineralization, the rate of which may be increased in

417 the presence of an alkaline environment (Gallagher et al., 2013). Copies of alkaline

418 phosphatase genes phoD and phoX are variable within bacterial genomes but the

419 majority of bacteria carry only 1 copy of phoD and 1 copy of phoX (Ragot et al., 2015,

420 2017). Conserved Acaryochloris and Rivularia species were particularly notable as they

421 encoded between 2–4 copies of phoX (Fig. 5), and carried all genes required for

422 phosphate transport (pstSCAB) (Fig. 5). Since the Cape Recife and Schoenmakerskop

423 stromatolites experience limited inorganic phosphate availability (Rishworth et al., 2016,

424 2018; Rishworth, Perissinotto, Bird, et al., 2017; Dodd et al., 2018), associated bacteria

425 may generate bioavailable phosphate from trapped sediments in biofilms resulting in

426 increased concentrations, similar to cyanobacteria living in phosphate-poor rivers

3- 2+ 427 (Wood et al., 2015). An increase in both PO4 and Ca can result in rapid precipitation

428 of apatite in marine phosphorites, freshwater lakes and in some soil bacteria (Danen-

429 Louwerse et al., 1995; Guang-Can et al., 2008; Cosmidis et al., 2015) and this could

430 account for the phosphatic deposits observed within the SA stromatolites (Buttner et al.,

431 2019, unpublished).

432

433 Finally, conserved Rivularia species in the middle and lower formations and other non-

434 conserved bacterial species harbored the 11 genes required for the transport (phnCDE)

435 and lysis (phnFGHIJKLM) of phosphonate compounds (Fig. 5) (Metcalf and Wanner,

436 1993). Phosphonates are characterized by the presence of a carbon-phosphorus bond

437 and are biosynthesized by many organisms (White and Metcalf, 2007). C-P lyase

438 (phnGHIJLKM) breaks the C-P bond within phosphonate substrates resulting in a

439 hydrocarbon and inorganic phosphate (White and Metcalf, 2007). The inorganic

440 phosphate may then be used within the bacterial cell or released aiding in accretion

441 through increased ion concentration (Rott et al., 2018). Conversely, phosphonates can

442 prevent precipitation of calcite by binding to crystal growth sites (Kan et al., 2005) and

443 the degradation of these compounds within stromatolite formations may prevent

444 chemical inhibition of stromatolite growth. Given the low availability of phosphate in

445 these systems, phosphonates may also provide an auxiliary source of bioavailable

446 phosphate. The potential for phosphonate degradation was also observed in 8 genomes

447 from the Shark Bay stromatolites (Wong et al., 2018). The conservation of phosphonate

448 metabolism across stromatolites would indicate that these bacterial processes may be

449 important within the generalized stromatolite system.

450

451 Archaeal genomes in Cape Recife and Schoenmakerskop stromatolites

452 Genome-resolved studies of hypersaline stromatolites in Shark Bay revealed that a

453 large proportion of the microbial community consisted of archaeal species (Wong et al.,

454 2017, 2018). Scrutiny of the contigs from Cape Recife and Schoenmakerskop showed

455 that none of the datasets comprised more than 1.5% archaeal genes. Assessment of

456 contigs clustered into the Archaeal kingdom bins from the upper and middle formations

457 showed little evidence for the presence of Archaea. However, two low-quality genomes

458 were obtained from the two lower formations SL_Arch_1 and RL_Arch_2, which were

459 classified as Woesearchaeales (order) and Nitrosoarchaeum (genus) respectively

460 (Table S3). Alignment of the 16S rRNA gene obtained from SL_Arch_1 (Table S3)

461 against the nr database showed that it shared the greatest sequence homology with an

462 uncultured archaeon clone GWA2 (87% identity). Similarly, analysis of the 16S rRNA

463 gene sequence from RL_Arch_2 (Table S3) indicated that it shared greatest sequence

464 homology with an uncultured archaeon clone 027 (99.79% identity). The coverage of

465 these genomes is among the lowest in each of the lower formation samples (Table S3)

466 and would suggest a low numerical abundance of these archaea. It is compelling that

467 the only Archaea detected in this study are derived from the formations most influenced

468 by saline waters, considering that the hypersaline stromatolites of Shark Bay comprise

469 several archaeal species (Wong et al., 2017, 2018), and it would be intriguing to identify

470 whether salinity has an effect on the Archaeal composition of stromatolite-associated

471 microbiota. However, as there are only two, low quality, non-conserved, relatively low-

472 abundance genomes derived from Archaea, it is likely that this kingdom is less

473 important in the stromatolites found at Cape Recife and Schoenmakerskop.

474

475 Key bacteria for stromatolite formation at Cape Recife and Schoenmakerskop

476 Several bacterial groups may be key players in the formation of stromatolites in Cape

477 Recife and Schoenmakerskop, with conserved bacterial species potentially acting as

478 major contributors (Table 1). There were distinct functional abilities between the

479 bacterial communities associated between the upper and middle formations. The

480 consortia in the upper formations were dominated by conserved Hydrococcus and

481 Acaryochloris species, which appeared to be capable of sulfate and sulfonate

482 metabolism, whilst the middle formations were dominated primarily by conserved

483 Rivularia species, which carry the genetic capacity to metabolize various forms of

484 phosphate substrates and potentially reduce nitrate. The lack of all genes required for

485 phosphonate metabolism in Rivularia species in Schoenmakerskop Middle 2 (SM2) was

486 unexpected but may be explained by the lower abundance of Rivularia in this sample,

487 resulting in lower quality of genome (Table S3). Similarly, Hydrococcus bin CSU_1_8

488 was of low quality (Table S3) and as a result appeared to be missing 3 of the genes

489 required.

490

491 We propose the redox potential and solubility index in these stromatolites is potentially

492 influenced by conserved bacteria generating sulfide (Acaryochloris and Hydrococcus

493 species) and ammonia ions (Rivularia species) for the rapid precipitation of carbonate

494 compounds and subsequent growth of stromatolite structures. The presence of

495 Microcoleus species (Phormidiaceae family) and an unidentified genus of

496 Phormidiaceae in both upper and middle formations is notable, as previous studies

497 have found that lamellar precipitation of carbonate occurs on the filaments of cultured

498 Phormidiaceae bacteria isolated from freshwater tufa structures (Payandi-Rolland et al.,

499 2019). We further propose an abundance of alkaline phosphatases in several

500 conserved species, as well as the abundance of genes associated with phosphonate

501 degradation in Rivularia species, may result in increased local inorganic phosphate

502 concentrations. This free inorganic phosphate may be incorporated into the phosphatic

503 crusts observed in these stromatolites (Buttner et al., 2019, unpublished).

504

505 The identification of conserved bacteria performing potentially important roles within

506 stromatolites of both Cape Recife and Schoenmakerskop may provide insight into what

507 cyanobacterial species may have played a key role in the formation of ancient

508 phosphatic stromatolites that formed in shallow marine and peritidal environments (Misi

509 and Kyle, 1994; Drummond et al., 2015; Buttner et al., 2019, unpublished; Shiraishi et

510 al., 2019). However, transcriptomics, nutrient uptake and additional hydrochemistry

511 would be required to determine if the proposed role of the conserved bacteria is valid.

512 Similarly, future studies will be needed to determine whether physical (e.g. flow rate) or

513 chemical factors (e.g. nutrient availability) result in our observed difference in functional

514 potential between the different formations.

515

516 EXPERIMENTAL PROCEDURES

517 Sample collection and DNA isolation. Samples were collected from

518 Schoenmakerskop (34°02'28.2"S 25°32'18.6"E) and Cape Recife (34°02'42.1"S

519 25°34'07.5"E) at low tide, from the water surface level. Samples approximately 1cm

520 deep were collected for 16S rRNA analysis in July 2019, from upper, middle and lower

521 stromatolite formations (Fig. 1) in triplicate, approximately 1 cm apart. Sample cores for

522 metagenomic shotgun sequencing were collected in April 2018, approximately 1cm

523 deep, from upper, middle and lower stromatolite formations. Additional samples were

524 collected in January 2018 from only the upper formations for metagenomic sequencing.

525 All samples were stored in RNAlater and flash-frozen until delivery to the lab at which

526 point, they were stored at -20 °C. DNA was extracted from ~1g of sample using Zymo

527 quick DNA Fecal/Soil Microbe Miniprep Kit (Zymo Research, Cat No. D6010) according

528 to the manufacturer's instructions.

529 Amplicon sequence analysis. Kapa HiFi Hotstart DNA polymerase (Roche, Cat No.

530 KK2500) was used to generate amplicon libraries of the V4-V5 region of the 16s rRNA

531 ribosomal subunit gene with the primer pair E517F (5’-GTAAGGTTCYTCGCGT-3’) and

532 E969-984 (5’-CAGCAGCCGCGGTAA-3’) (Matcher et al., 2011) from triplicate samples

533 using the following cycling parameters: Initial denaturation at 98 °C; 5 cycles (98 °C for

534 45 seconds, 45 °C for 45 seconds and 72 °C for 1 minute); 18 cycles (98 °C for 45

535 seconds, 50 °C for 30 seconds and 72 °C for 1 minute); final elongation step at 72 °C

536 for 5 minutes. PCR products were purified using the Bioline Isolate II PCR and Gel kit

537 (Bioline, Cat. No. BIO-52060). Samples were sequenced using the Illumina Miseq

538 platform. Amplicon library datasets were processed and curated using the Mothur

539 software platform (Schloss et al., 2009). Sequences shorter than 200 nucleotides or

540 containing ambiguous bases or homopolymeric runs greater than 7 were discarded.

541 Sequences were classified using Naive-Bayesian classifier against the Silva database

542 (v132) and VSEARCH software (Rognes et al., 2016) was used to remove chimeras.

543 Reads that were 97% similar were combined into operational taxonomic units (OTUs)

544 using the Opticlust method (Westcott and Schloss, 2017). OTU abundance values were

545 converted to relative values and the dataset was then transformed by square root and

546 statistically analyzed using the Primer-e (V7) software package (Gorley and Clarke,

547 2015). Reads were not classified against the Greengenes database, as the creators of

548 Mothur warn against this practice, citing poor alignment quality in the variable regions.

549 Metagenomic binning. As 16S rRNA sequence analysis indicated that there was no

550 statistical difference between triplicate samples, and that sample regions are therefore

551 homogenous, a single sample from sites CRU, RU, RM1, RM2, RL, CSU, SU, SM1,

552 SM2 and SL (Fig.1) collected in January and April 2018 was used to prepare shotgun

553 DNA libraries that were sequenced using IonTorrent Ion P1.1.17 Chip technology

554 (Central Analytical Facilities, Stellenbosch University, South Africa). Adapters were

555 trimmed and trailing bases (15 nts) were iteratively removed if the average quality score

556 for the last 30 nts was lower than 16, which resulted in approximately 30–45 million

557 reads per sample. Resultant metagenomic datasets were assembled into contiguous

558 sequences (contigs) with SPAdes version 3.12.0 (Bankevich et al., 2012) using the --

559 iontorrent and --only-assembler options with kmer values of 21,33,55,77,99,127. The --

560 iontorrent option enables a read error correction step performed by IonHammer,

561 correcting homopolymeric runs inherent to IonTorrent sequencing chemistry (Ershov et

562 al., 2019).

563 Quantification of metabolism-associated genes. All raw data, count tables and

564 scripts used to process the data can be accessed here:

565 https://github.com/samche42/Conserved_Stromatolite_bacteria_manuscript.git

566 Contigs in each of the 10 samples were clustered into kingdom bins using Autometa

567 (Miller et al., 2019). Genes on all contigs within these bins were identified using Prodigal

568 v.2.6.3 (Hyatt et al., 2010). Genes were then annotated against the KEGG database

569 using kofamscan (Aramaki et al., 2019) with output in mapper format. Coverage

570 corrected KO annotation counts were collected using kegg_parser.py found in the

571 GitHub repo listed above. Briefly, a dictionary of collection sites was made with all KO

572 numbers supplied in a KEGG annotation query list (i.e. functional group). Kofamscan

573 output per collection site was parsed and each time an entry matching that within the

574 query list was found, a count for that annotation was increased by the coverage of the

575 contig on which it was located, resulting in a coverage-corrected count table of

576 functional group KO annotations per collection site. In order to assess differences

577 between collection sites, these values were transformed relative to average coverage

578 per sample (weighted by contig length) using weighted_contig_coverage_calculator.py

579 (see GitHub repo) according to Eq.1, where i is contig length and j is contig coverage.

580 The abundances were then transformed using log2 (Fig. 3).

581 Equation 1. Contig length weighted coverage = ∑( (i / ∑(i)) x j )

582 Heatmaps for visualization of the data were created in R using the tidyr, ggplot2 and

583 viridis packages. Scripts can be found in the GitHub repo.

584 Clustering of metagenomic data in genome bins. Contigs were then clustered in

585 putative genomic bins using Autometa

586 (https://bitbucket.org/jason_c_kwan/autometa/src/master/, Master branch, commit:

587 a344c28) (Miller et al., 2019). Bins were manually curated and resulted in 183 genomic

588 bins. CheckM (Parks et al., 2015) was used to assess bin purity and completion using

589 default settings (Table S3). Approximately 37 genomes were of high quality, 75 were of

590 medium quality and 71 were of low quality, in accordance with MIMAG standards

591 defined by (Bowers et al., 2017) (Table S3).

592 Identification of conserved taxa. Conservation of bacterial taxa was calculated using

593 average nucleotide identities (ANIs) of all genomic bins, which were calculated in a

594 pairwise manner using FastANI (Jain et al., 2018). All genomic pairs sharing > 97% ANI

595 were subset and considered conserved taxa. Percentage of mapped regions used to

596 calculate ANI is reported in Table S2. Percentage of mapped regions is calculated as

597 the relative number of bidirectional fragments mapping to the total number of query

598 fragments.

599 Genome taxonomic classification. Genomes were classified using the standalone

600 GTDB-Tk tool (version 0.3.2) using the classify workflow and Genome Taxonomy

601 Database version 89 (Parks et al., 2019). The tool is unable to classify genomes less

602 than 10% complete, as indicated in Table S3. The GTDB-Tk tool uses the Genome

603 Taxonomy Database as a reference for classification, which is based on phylogeny

604 inferred from concatenated protein alignments. This approach enabled the removal of

605 polyphyletic groups and assignment of taxonomy from evolutionary divergence. The

606 resulting taxonomy incorporates substantial changes in comparison to the NCBI

607 taxonomy (Parks et al., 2018). Equivalent NCBI taxonomic classifications have been

608 provided for clarity in Table S3.

609 Genome taxonomic clustering. Phylogeny of bacterial genomes was inferred using

610 JolyTree (Criscuolo, 2019) with a sketch size of 10 000 (Fig. 5 and Fig. S1). JolyTree

611 infers phylogeny through computation of dissimilarity of kmer sketches, which is then

612 transformed for the estimation of substitution events of the genomes’ evolution

613 (Criscuolo, 2019).

614 Genome annotation. Manually curated putative genomic bins were annotated using

615 Prokka version 1.13 (Seemann, 2014), with GenBank compliance enabled. Protein-

616 coding amino-acid sequences from genomic bins were annotated against the KEGG

617 database using kofamscan (Aramaki et al., 2019) with output in mapper format. KEGG

618 orthologs were counted and processed as performed quantification of metabolism-

619 associated genes using kegg_parser.py. Relative percentage of genes in individual

620 genomes was calculated by dividing the contig-coverage corrected gene abundance per

621 gene by the total contig-coverage corrected gene abundance for the sample from which

622 it was binned (Fig. 5 and Fig. S1). E.g. The contig-coverage corrected gene abundance

623 of phoU in genome bin CRU1_1 was 3.52, and the total contig-coverage corrected gene

624 abundance of phoU in the CRU metagenome sample was 29.90253 (See dataset

625 Calculating_rel_perc.xlsx in GitHub repo). Therefore, the relative abundance of phoU in

626 genome bin CRU1_1 was: 3.52/29.90 X 100 = 11.79% of total sample gene abundance.

627 Archaeal genome binning and identification. Contigs are classified within kingdom

628 bins during Autometa binning. All contigs classified within the Archaea kingdom were

629 given putative taxonomic assignments based on single-copy markers and clustered into

630 bins. No bins could be acquired from the upper and middle formations from either

631 collection site due to few or no contigs being of archaeal origin. Two genomes were

632 obtained from the lower formation of each site respectively (Table S3). Genome quality

633 was assessed using CheckM as described for bacterial bins. 16S rRNA and 23S rRNA

634 sequences were extracted from each genome using barrnap 0.9

635 (https://github.com/tseemann/barrnap) using the archaeal databases (23S: SILVA-LSU-

636 Arc, 16S: RF01959) as reference. These sequences were then aligned against the nr

637 database using BLASTn (Johnson et al., 2008) for putative identification.

638 Data availability. Raw 16S rRNA gene amplicon sequence files were uploaded to the

639 NCBI sequence read archive (SRA) database in BioProject PRJNA574289. Raw reads

640 and binned genomes will be deposited in GenBank and respective accession numbers

641 will be included in the accepted version of this manuscript.

642 ACKNOWLEDGEMENTS

643 The authors acknowledge Caro Damarjanan who conducted a pilot study that informed

644 this work. We also wish to thank Karthik Anantharaman (University of Wisconsin-

645 Madison) for his helpful critique.

646

647 The authors acknowledge funding from the Gordon and Betty Moore Foundation (Grant

648 number 6920) (awarded to R.A.D and J.C.K.), and grants awarded to R.A.D by the

649 South African National Research Foundation (UID: 87583 and 109680). Development of

650 Autometa in J.C.K.’s laboratory and contributions by E.R.R. were supported by the U.S.

651 National Science Foundation (DBI-1845890). This research was performed in part using

652 the computer resources and assistance of the UW-Madison Center for High Throughput

653 Computing (CHTC) in the Department of Computer Sciences. The CHTC is supported

654 by UW-Madison, the Advanced Computing Initiative, the Wisconsin Alumni Research

655 Foundation, Wisconsin Institutes for Discovery, and the National Science Foundation

656 and is an active member of the Open Science Grid, which is supported by the National

657 Science Foundation and the U.S. Department of Energy’s Office of Science. The

658 authors also acknowledge the Center for High Performance Computing (CHPC, South

659 Africa) for providing computing facilities for bioinformatics data analysis.

660

661 The authors declare no conflict of interests.

662

663

664 REFERENCES

665 Aguilar-Barajas, E., Díaz-Pérez, C., Ramírez-Díaz, M.I., Riveros-Rosas, H., and

666 Cervantes, C. (2011) Bacterial transport of sulfate, molybdate, and related oxyanions.

667 Biometals 24: 687–707.

668 Allen, J.F. (2016) A Proposal for Formation of Archaean stromatolites before the advent

669 of oxygenic photosynthesis. Front Microbiol 7: 1784.

670 Aramaki, T., Blanc-Mathieu, R., Endo, H., Ohkubo, K., Kanehisa, M., Goto, S., and

671 Ogata, H. (2019) KofamKOALA: KEGG ortholog assignment based on profile HMM and

672 adaptive score threshold. bioRxiv. DOI: https://doi.org/10.1101/602110

673 Awramik, S.M. (1992) The History and Significance of Stromatolites. In, Schidlowski,M.,

674 Golubic,S., Kimberley,M.M., McKirdy,D.M., and Trudinger,P.A. (eds), Early Organic

675 Evolution: Implications for Mineral and Energy Resources. Berlin, Heidelberg: Springer

676 Berlin Heidelberg, pp. 435–449.

677 Babilonia, J., Conesa, A., Casaburi, G., Pereira, C., Louyakis, A.S., Reid, R.P., and

678 Foster, J.S. (2018) Comparative metagenomics provides insight into the ecosystem

679 functioning of the Shark Bay Stromatolites, Western Australia. Front Microbiol 9: 1359.

680 Balci, N., Gunes, Y., Sertcetin, E., and Kurt, M.A. (2018) Non-cyanobacterial community

681 shape the stromatolites of Lake Salda, SW, Turkey. American Geophysical Union, Fall

682 Meeting 2018, abstract #B43J-2972.

683 Bankevich, A., Nurk, S., Antipov, D., Gurevich, A.A., Dvorkin, M., Kulikov, A.S., et al.

684 (2012) SPAdes: A new genome assembly algorithm and its applications to single-cell

685 sequencing. J Comput Biol 19: 455–477.

686 Bontognali, T.R.R., Sessions, A.L., Allwood, A.C., Fischer, W.W., Grotzinger, J.P.,

687 Summons, R.E., and Eiler, J.M. (2012) Sulfur isotopes of organic matter preserved in

688 3.45-billion-year-old stromatolites reveal microbial metabolism. Proc Natl Acad Sci U S

689 A 109: 15146–15151.

690 Bowers, R.M., Kyrpides, N.C., Stepanauskas, R., Harmon-Smith, M., Doud, D., Reddy,

691 T.B.K., et al. (2017) Minimum information about a single amplified genome (MISAG)

692 and a metagenome-assembled genome (MIMAG) of bacteria and archaea. Nat

693 Biotechnol 35: 725–731.

694 Casaburi, G., Duscher, A.A., Pamela Reid, R., and Foster, J.S. (2016) Characterization

695 of the stromatolite microbiome from Little Darby Island, The Bahamas using predictive

696 and whole shotgun metagenomic analysis. Environ Microbiol 18: 1452–1469.

697 Centeno, C.M., Mejía, O., and Falcón, L.I. (2016) Habitat conditions drive phylogenetic

698 structure of dominant bacterial phyla of microbialite communities from different locations

699 in Mexico. Rev Biol Trop 64: 1057–1065.

700 Chaumeil, P.-A., Mussig, A.J., Hugenholtz, P., and Parks, D.H. (2019) GTDB-Tk: A

701 toolkit to classify genomes with the Genome Taxonomy Database. Bioinformatics.

702 btz848.

703 Cosmidis, J., Benzerara, K., Guyot, F., Skouri-Panet, F., Duprat, E., Férard, C., et al.

704 (2015) Calcium-phosphate biomineralization induced by alkaline phosphatase activity in

705 Escherichia coli: Localization, kinetics, and potential signatures in the fossil record.

706 Front Earth Sci 3: 84.

707 Criscuolo, A. (2019) A fast alignment-free bioinformatics procedure to infer accurate

708 distance-based phylogenetic trees from genome assemblies. Riogrande Odontol 5:

709 e36178.

710 Danen-Louwerse, H.J., Lijklema, L., and Coenraats, M. (1995) Coprecipitation of

711 phosphate with calcium carbonate in Lake Veluwe. Water Res 29: 1781–1785.

712 Dodd, C., Anderson, C.R., Perissinotto, R., du Plooy, S.J., and Rishworth, G.M. (2018)

713 Hydrochemistry of peritidal stromatolite pools and associated freshwater inlets along the

714 Eastern Cape coast, South Africa. Sediment Geol 373: 163–179.

715 Drummond, J.B.R., Pufahl, P.K., Porto, C.G., and Carvalho, M. (2015) Neoproterozoic

716 peritidal phosphorite from the Sete Lagoas Formation (Brazil) and the Precambrian

717 phosphorus cycle. Sedimentology 62: 1978–2008.

718 Dupraz, C., Reid, R.P., Braissant, O., Decho, A.W., Norman, R.S., and Visscher, P.T.

719 (2009) Processes of carbonate precipitation in modern microbial mats. Earth-Sci Rev

720 96: 141–162.

721 Dupraz, C., Visscher, P.T., Baumgartner, L.K., and Reid, R.P. (2004) Microbe-mineral

722 interactions: Early carbonate precipitation in a hypersaline lake (Eleuthera Island,

723 Bahamas). Sedimentology 51: 745–765.

724 Ellis, H.R. (2011) Mechanism for sulfur acquisition by the alkanesulfonate

725 monooxygenase system. Bioorg Chem 39: 178–184.

726 Ershov, V., Tarasov, A., Lapidus, A., and Korobeynikov, A. (2019) IonHammer:

727 Homopolymer-space hamming clustering for IonTorrent read error correction. J Comput

728 Biol 26: 124–127.

729 Gallagher, K.L., Braissant, O., Kading, T.J., Dupraz, C., and Visscher, P.T. (2013)

730 Phosphate-related artifacts in carbonate mineralization experiments. J Sediment Res

731 83: 37–49.

732 Gallagher, K.L., Kading, T.J., Braissant, O., Dupraz, C., and Visscher, P.T. (2012)

733 Inside the alkalinity engine: The role of electron donors in the organomineralization

734 potential of sulfate-reducing bacteria. Geobiology 10: 518–530.

735 Gleeson, D.B., Wacey, D., Waite, I., O’Donnell, A.G., and Kilburn, M.R. (2016)

736 Biodiversity of living, non-marine, thrombolites of Lake Clifton, Western Australia.

737 Geomicrobiol J 33: 850–859.

738 Gorley, R.N. and Clarke, K.R. (2015) PRIMER v7: User Manual/Tutorial, Massey

739 University, Albany Campus.

740 Griffith, J.C. (2016) Insights into the soil microbial communities in New Zealand

741 indigenous tussock grasslands. (Thesis, Doctor of Philosophy). University of Otago.

742 Retrieved from http://hdl.handle.net/10523/6906

743 Guang-Can, T.A.O., Shu-Jun, T., Miao-Ying, C.A.I., and Guang-Hui, X.I.E. (2008)

744 Phosphate-solubilizing and-mineralizing abilities of bacteria isolated from soils.

745 Pedosphere 18: 515–523.

746 Herrmann, M., Wegner, C.-E., Taubert, M., Geesink, P., Lehmann, K., Yan, L., et al.

747 (2019) Predominance of Cand. Patescibacteria in groundwater is caused by their

748 preferential mobilization from soils and flourishing under oligotrophic conditions. Front

749 Microbiol 10: 1407.

750 Hyatt, D., Chen, G.-L., Locascio, P.F., Land, M.L., Larimer, F.W., and Hauser, L.J.

751 (2010) Prodigal: Prokaryotic gene recognition and translation initiation site identification.

752 BMC Bioinformatics 11: 119.

753 Jain, C., Rodriguez-R, L.M., Phillippy, A.M., Konstantinidis, K.T., and Aluru, S. (2018)

754 High throughput ANI analysis of 90K prokaryotic genomes reveals clear species

755 boundaries. Nat Commun 9: 5114.

756 Johnson, M., Zaretskaya, I., Raytselis, Y., Merezhuk, Y., McGinnis, S., and Madden,

757 T.L. (2008) NCBI BLAST: A better web interface. Nucleic Acids Res 36: W5–9.

758 Kahnert, A., Vermeij, P., Wietek, C., James, P., Leisinger, T., and Kertesz, M.A. (2000)

759 The ssu locus plays a key role in organosulfur metabolism in Pseudomonas putida S-

760 313. J Bacteriol 182: 2869–2878.

761 Kan, A.T., Fu, G., and Tomson, M.B. (2005) Adsorption and precipitation of an

762 aminoalkylphosphonate onto calcite. J Colloid Interface Sci 281: 275–284.

763 Keegan, K.P., Glass, E.M., and Meyer, F. (2016) MG-RAST, a metagenomics service

764 for analysis of microbial community structure and function. Microbial Environmental

765 Genomics (MEG) 1399: 207–233.

766 Khodadad, C.L.M. and Foster, J.S. (2012) Metagenomic and metabolic profiling of

767 nonlithifying and lithifying stromatolitic mats of Highborne Cay, the Bahamas. PLoS One

768 7: e38229.

769 Konopacka-Łyskawa, D., Kościelska, B., Karczewski, J., and Gołąbiewska, A. (2017)

770 The influence of ammonia and selected amines on the characteristics of calcium

771 carbonate precipitated from calcium chloride solutions via carbonation. Mater Chem

772 Phys 193: 13–18.

773 Kurth, D., Amadio, A., Ordoñez, O.F., Albarracín, V.H., Gärtner, W., and Farías, M.E.

774 (2017) Arsenic metabolism in high altitude modern stromatolites revealed by

775 metagenomic analysis. Sci Rep 7: 1024.

776 Lee, Y.S. and Park, W. (2019) Enhanced calcium carbonate-biofilm complex formation

777 by alkali-generating Lysinibacillus boronitolerans YS11 and alkaliphilic Bacillus sp.

778 AK13. AMB Express 9: 49.

779 Macintyre, I.G., Prufert-Bebout, L., and Reid, R.P. (2000) The role of endolithic

780 cyanobacteria in the formation of lithified laminae in Bahamian stromatolites.

781 Sedimentology 47: 915–921.

782 Matcher, G.F., Dorrington, R.A., Henninger, T.O., and Froneman, P.W. (2011) Insights

783 into the bacterial diversity in a freshwater-deprived permanently open Eastern Cape

784 estuary, using 16S rRNA pyrosequencing analysis. Water SA 37.3: 381 - 390.

785 Medina-Chávez, N.O., Viladomat-Jasso, M., Olmedo-Álvarez, G., Eguiarte, L.E., Souza,

786 V., and De la Torre-Zavala, S. (2019) Diversity of Archaea domain in Cuatro Cienegas

787 Basin: Archaean Domes. bioRxiv 766709. DOI: https://doi.org/10.1101/766709

788 Metcalf, W.W. and Wanner, B.L. (1993) Evidence for a fourteen-gene, phnC to phnP

789 locus for phosphonate metabolism in Escherichia coli. Gene 129: 27–32.

790 Miller, I.J., Rees, E.R., Ross, J., Miller, I., Baxa, J., Lopera, J., et al. (2019) Autometa:

791 Automated extraction of microbial genomes from individual shotgun metagenomes.

792 Nucleic Acids Res 47: e57.

793 Misi, A. and Richard Kyle, J. (1994) Upper Proterozoic carbonate stratigraphy,

794 diagenesis, and stromatolitic phosphorite formation, Irece Basin, Bahia, Brazil. J

795 Sediment Res 64: 299–310.

796 Mobberley, J.M., Khodadad, C.L.M., Visscher, P.T., Reid, R.P., Hagan, P., and Foster,

797 J.S. (2015) Inner workings of thrombolites: Spatial gradients of metabolic activity as

798 revealed by metatranscriptome profiling. Sci Rep 5: 12601.

799 Moreno-Vivián, C., Cabello, P., Martínez-Luque, M., Blasco, R., and Castillo, F. (1999)

800 Prokaryotic nitrate reduction: Molecular properties and functional distinction among

801 bacterial nitrate reductases. J Bacteriol 181: 6573–6584.

802 Nutman, A.P., Bennett, V.C., Friend, C.R.L., Van Kranendonk, M.J., and Chivas, A.R.

803 (2016) Rapid emergence of life shown by discovery of 3,700-million-year-old microbial

804 structures. Nature 537: 535–538.

805 Pace, A., Bourillot, R., Bouton, A., Vennin, E., Braissant, O., Dupraz, C., et al. (2018)

806 Formation of stromatolite lamina at the interface of oxygenic-anoxygenic

807 photosynthesis. Geobiology 16: 378–398.

808 Parks, D.H., Chuvochina, M., Chaumeil, P.-A., Rinke, C., Mussig, A.J., and Hugenholtz,

809 P. (2019) Selection of representative genomes for 24,706 bacterial and archaeal

810 species clusters provide a complete genome-based taxonomy. bioRxiv 771964. DOI:

811 https://doi.org/10.1101/771964

812 Parks, D.H., Chuvochina, M., Waite, D.W., Rinke, C., Skarshewski, A., Chaumeil, P.-A.,

813 and Hugenholtz, P. (2018) A standardized bacterial taxonomy based on genome

814 phylogeny substantially revises the tree of life. Nat Biotechnol 36: 996–1004.

815 Parks, D.H., Imelfort, M., Skennerton, C.T., Hugenholtz, P., and Tyson, G.W. (2015)

816 CheckM: Assessing the quality of microbial genomes recovered from isolates, single

817 cells, and metagenomes. Genome Res 25: 1043–1055.

818 Payandi-Rolland, D., Roche, A., Vennin, E., Visscher, P.T., Amiotte-Suchet, P.,

819 Thomas, C., and Bundeleva, I.A. (2019) Carbonate precipitation in mixed cyanobacterial

820 biofilms forming freshwater microbial tufa. Minerals 9: 409.

821 Perissinotto, R., Bornman, T.G., Steyn, P.-P., Miranda, N.A.F., Dorrington, R.A.,

822 Matcher, G.F., et al. (2014) Tufa stromatolite ecosystems on the South African south

823 coast. S Afr J Sci 110: 01–08.

824 Ragot, S.A., Kertesz, M.A., and Bünemann, E.K. (2015) phoD Alkaline phosphatase

825 gene diversity in soil. Appl Environ Microbiol 81: 7281–7289.

826 Ragot, S.A., Kertesz, M.A., Mészáros, É., Frossard, E., and Bünemann, E.K. (2017) Soil

827 phoD and phoX alkaline phosphatase gene diversity responds to multiple environmental

828 factors. FEMS Microbiol Ecol 93: fiw212

829 Reid, R.P., Visscher, P.T., Decho, A.W., Stolz, J.F., Bebout, B.M., Dupraz, C., et al.

830 (2000) The role of microbes in accretion, lamination and early lithification of modern

831 marine stromatolites. Nature 406: 989–992.

832 Rishworth, G.M., Cawthra, H.C., Dodd, C., and Perissinotto, R. (2019) Peritidal

833 stromatolites as indicators of stepping-stone freshwater resources on the Palaeo-

834 Agulhas Plain landscape. Quat Sci Rev 105704.

835 Rishworth, G.M., van Elden, S., Perissinotto, R., Miranda, N.A.F., Steyn, P.-P., and

836 Bornman, T.G. (2016) Environmental influences on living marine stromatolites: Insights

837 from benthic microalgal communities. Environ Microbiol 18: 503–513.

838 Rishworth, G.M., Perissinotto, R., Bird, M.S., and Pelletier, N. (2018) Grazer responses

839 to variable macroalgal resource conditions facilitate habitat structuring. R Soc Open Sci

840 5: 171428.

841 Rishworth, G.M., Perissinotto, R., Bird, M.S., Strydom, N.A., Peer, N., Miranda, N.A.F.,

842 and Raw, J.L. (2017) Non-reliance of metazoans on stromatolite-forming microbial mats

843 as a food resource. Sci Rep 7: 42614.

844 Rishworth, G.M., Perissinotto, R., Bornman, T.G., and Lemley, D.A. (2017) Peritidal

845 stromatolites at the convergence of groundwater seepage and marine incursion:

846 Patterns of salinity, temperature and nutrient variability. J Mar Syst 167: 68–77.

847 Rodriguez-Navarro, C., Rodriguez-Gallego, M., Ben Chekroun, K., and Gonzalez-

848 Muñoz, M.T. (2003) Conservation of ornamental stone by Myxococcus xanthus-induced

849 carbonate biomineralization. Appl Environ Microbiol 69: 2182–2193.

850 Rognes, T., Flouri, T., Nichols, B., Quince, C., and Mahé, F. (2016) VSEARCH: A

851 versatile open source tool for metagenomics. PeerJ 4: e2584.

852 Rott, E., Steinmetz, H., and Metzger, J.W. (2018) Organophosphonates: A review on

853 environmental relevance, biodegradability and removal in wastewater treatment plants.

854 Sci Total Environ 615: 1176–1191.

855 Ruvindy, R., White, R.A., 3rd, Neilan, B.A., and Burns, B.P. (2016) Unravelling core

856 microbial metabolisms in the hypersaline microbial mats of Shark Bay using high-

857 throughput metagenomics. ISME J 10: 183–196.

858 Santos, A.A., Venceslau, S.S., Grein, F., Leavitt, W.D., Dahl, C., Johnston, D.T., and

859 Pereira, I.A.C. (2015) A protein trisulfide couples dissimilatory sulfate reduction to

860 energy conservation. Science 350: 1541–1545.

861 Schloss, P.D., Westcott, S.L., Ryabin, T., Hall, J.R., Hartmann, M., Hollister, E.B., et al.

862 (2009) Introducing mothur: Open-source, platform-independent, community-supported

863 software for describing and comparing microbial communities. Appl Environ Microbiol

864 75: 7537–7541.

865 Seemann, T. (2014) Prokka: Rapid prokaryotic genome annotation. Bioinformatics 30:

866 2068–2069.

867 Shiraishi, F., Ohnishi, S., Hayasaka, Y., Hanzawa, Y., Takashima, C., Okumura, T., and

868 Kano, A. (2019) Potential photosynthetic impact on phosphate stromatolite formation

869 after the Marinoan glaciation: Paleoceanographic implications. Sediment Geol 380: 65–

870 82.

871 Smith, A., Cooper, A., Misra, S., Bharuth, V., Guastella, L., and Botes, R. (2018) The

872 extant shore platform stromatolite (SPS) facies association: A glimpse into the

873 Archean? Biogeosciences 15: 2189–2203.

874 Soo, R.M., Hemp, J., Parks, D.H., Fischer, W.W., and Hugenholtz, P. (2017) On the

875 origins of oxygenic photosynthesis and aerobic respiration in Cyanobacteria. Science

876 355: 1436–1440.

877 Stasi, R., Neves, H.I., and Spira, B. (2019) Phosphate uptake by the phosphonate

878 transport system PhnCDE. BMC Microbiol 19: 79.

879 Stüeken, E.E. (2016) Nitrogen in ancient mud: A biosignature? Astrobiology 16: 730–

880 735.

881 Thomazo, C., Ader, M., and Philippot, P. (2011) Extreme 15N-enrichments in 2.72-Gyr-

882 old sediments: Evidence for a turning point in the nitrogen cycle. Geobiology 9: 107–

883 120.

884 Visscher, P.T., Pamela Reid, R., and Bebout, B.M. (2000) Microscale observations of

885 sulfate reduction: Correlation of microbial activity with lithified micritic laminae in modern

886 marine stromatolites. Geology 28: 919–922.

887 Visscher, P.T. and Stolz, J.F. (2005) Microbial mats as bioreactors: Populations,

888 processes, and products. In, Noffke,N. (ed), Geobiology: Objectives, Concepts,

889 Perspectives. Amsterdam: Elsevier, pp. 87–100.

890 Warden, J.G., Casaburi, G., Omelon, C.R., Bennett, P.C., Breecker, D.O., and Foster,

891 J.S. (2016) Characterization of microbial mat microbiomes in the modern thrombolite

892 ecosystem of Lake Clifton, Western Australia using shotgun metagenomics. Front

893 Microbiol 7: 1064.

894 Wei, S., Cui, H., Jiang, Z., Liu, H., He, H., and Fang, N. (2015) Biomineralization

895 processes of calcite induced by bacteria isolated from marine sediments. Braz J

896 Microbiol 46: 455–464.

897 Westcott, S.L. and Schloss, P.D. (2017) OptiClust, an improved method for assigning

898 amplicon-based sequence data to operational taxonomic units. mSphere 2: e00073-17.

899 White, A.K. and Metcalf, W.W. (2007) Microbial metabolism of reduced phosphorus

900 compounds. Annu Rev Microbiol 61: 379–400.

901 White, R.A., Chan, A.M., Gavelis, G.S., Leander, B.S., Brady, A.L., Slater, G.F., et al.

902 (2016) Metagenomic analysis suggests modern freshwater microbialites harbor a

903 distinct core microbial community. Front Microbiol 6: 1531.

904 Witze, A. (2016) Claims of Earth’s oldest fossils tantalize researchers. Nature News.

905 DOI:10.1038/nature.2016.20506

906 Wong, H.L., Visscher, P.T., White, R.A., III, Smith, D.-L., Patterson, M.M., and Burns,

907 B.P. (2017) Dynamics of archaea at fine spatial scales in Shark Bay mat microbiomes.

908 Sci Rep 7: 46160.

909 Wong, H.L., White, R.A., Visscher, P.T., Charlesworth, J.C., Vázquez-Campos, X., and

910 Burns, B.P. (2018) Disentangling the drivers of functional complexity at the

911 metagenomic level in Shark Bay microbial mat microbiomes. ISME J 12: 2619–2639.

912 Wood, S.A., Depree, C., Brown, L., McAllister, T., and Hawes, I. (2015) Entrapped

913 sediments as a source of phosphorus in epilithic cyanobacterial proliferations in low

914 nutrient rivers. PLoS One 10: e0141063.

915 Yang, J., Junium, C.K., Grassineau, N.V., Nisbet, E.G., Izon, G., Mettam, C., et al.

916 (2019) Ammonium availability in the Late Archaean nitrogen cycle. Nature Geoscience

917 12: 553–557.

918 Zaheer, R., Morton, R., Proudfoot, M., Yakunin, A., and Finan, T.M. (2009) Genetic and

919 biochemical properties of alkaline phosphatases phoX family protein found in many

920 bacteria. Environ Microbiol 11: 1572–1587.

921

922 TABLE AND FIGURE LEGENDS

923 Figure 1. Stromatolites were collected from four different points at two different sites

924 along the SA coastline. (A) Aerial view of sampling locations within the

925 Schoenmakerskop site and (B) the Cape Recife site. Contour lines represent elevation

926 at 5cm increments. Samples were collected from upper, middle and lower formations.

927 The boundaries of which are indicated with a dotted blue line. (C - D) A 3-dimensional

928 rendering of the sample sites at Schoenmakerskop and Cape Recife respectively

929 constructed by a 3rd party (Caelum Technologies) using a combination of drone-based

930 photogrammetry and geographic mapping using differential GPS. Samples were

931 collected from upper, middle and lower formations; the boundaries of which are

932 indicated with a dotted blue line. Abbreviations are as follows: CSU: Schoenmakerskop

933 Upper Jan SU: Schoenmakerskop Upper April, CRU: Cape Recife Upper (Jan), RU:

934 Cape Recife Upper (April), SM1: Schoenmakerskop Middle 1 (April), SM2:

935 Schoenmakerskop Middle 2 (April), RM1: Cape Recife Middle 1 (April), RM2: Cape

936 Recife Middle 1 (April), SL: Schoenmakerskop Lower (April), RL: Cape Recife Lower

937 (April).

938

939 Figure 2. Distribution and abundance of bacterial taxa in stromatolite formations based

940 on 16S rRNA gene fragment amplicon libraries. (A) Phylogenetic classification and

941 average relative abundance (n=3) of dominant phyla in different sample sites indicated

942 that all stromatolite samples are dominated by Cyanobacteria, Bacteroidetes, Alpha-

943 and Gammaproteobacteria. (B) OTU abundance was used to cluster stromatolite

944 biological replicates using Bray-Curtis non-dimensional scaling and showed statistically

945 significant clustering of replicate samples from each of the sampled regions. Samples

946 were isolated from upper (green), middle (red) and lower (blue) stromatolites formations

947 in Cape Recife (triangles) and Schoenmakerskop (circles).

948

949 Figure 3. Summary of phosphate, nitrogen and sulfate transport and metabolism genes

950 in overall metagenomic data from stromatolites in Schoenmakerskop and Cape Recife

951 upper, middle and lower formations respectively. Gene abundance is expressed relative

952 to the length-weighted average contig coverage per sample and transformed using log2

953 scaling. Note: On the color scale, grey indicates genes which were not detected in the

954 respective sample. Abbreviations are as follows: CSU: Schoenmakerskop Upper Jan

955 SU: Schoenmakerskop Upper April, CRU: Cape Recife Upper (Jan), RU: Cape Recife

956 Upper (April), SM1: Schoenmakerskop Middle 1 (April), SM2: Schoenmakerskop Middle

957 2 (April), RM1: Cape Recife Middle 1 (April), RM2: Cape Recife Middle 1 (April), SL:

958 Schoenmakerskop Lower (April), RL: Cape Recife Lower (April).

959

960 Figure 4. Taxonomic classification of putative genome bins in stromatolites collected

961 from upper/inflow, middle and lower/marine formations of Schoenmakerskop and Cape

962 Recife. Coverage per genome has been used as a proxy for abundance and used to

963 scale the size of individual genome bars, expressed as a percentage of the sum of all

964 genome bin coverages. (A) The coverage of conserved (diagonal lines) and non-

965 conserved (solid colors) bacterial genomes. Taxonomic classification of each genome is

966 indicated by color. (B). Taxonomic classifications of conserved bacterial species in each

967 of the samples, generated with GTDB-Tk (Chaumeil et al., 2019).

968

969 Figure 5. Summary of phosphate, nitrogen and sulfate transport and metabolism genes

970 in conserved genomes from bacteria associated with stromatolites in Schoenmakerskop

971 and Cape Recife upper, middle and lower formations respectively. Relative gene

972 abundance per sample is indicated using a gradient color scale. Relative gene

973 abundance was calculated by dividing coverage-corrected gene abundance in each

974 genome by the coverage-corrected gene abundance from the metagenomic sample

975 from which it came. CSU: Schoenmakerskop Upper Jan SU: Schoenmakerskop Upper

976 April, CRU: Cape Recife Upper (Jan), RU: Cape Recife Upper (April), SM1:

977 Schoenmakerskop Middle 1 (April), SM2: Schoenmakerskop Middle 2 (April), RM1:

978 Cape Recife Middle 1 (April), RM2: Cape Recife Middle 1 (April), SL: Schoenmakerskop

979 Lower (April), RL: Cape Recife Lower (April).

980

981 Figure S1. Relative abundance of phosphate, nitrogen and sulfate transport and

982 metabolism genes per genome relative to total gene abundance per sample from

983 Schoenmakerskop and Cape Recife upper, middle and lower formations. Relative gene

984 abundance per sample is indicated using a gradient color scale. Relative gene

985 abundance was calculated by dividing coverage-corrected gene abundance in each

986 genome by the coverage-corrected gene abundance from the metagenomic sample

987 from which it came. CSU: Schoenmakerskop Upper Jan SU: Schoenmakerskop Upper

988 April, CRU: Cape Recife Upper (Jan), RU: Cape Recife Upper (April), SM1:

989 Schoenmakerskop Middle 1 (April), SM2: Schoenmakerskop Middle 2 (April), RM1:

990 Cape Recife Middle 1 (April), RM2: Cape Recife Middle 1 (April), SL: Schoenmakerskop

991 Lower (April), RL: Cape Recife Lower (April).

992

993 Table 1. Conserved bacterial species (ANI > 97%) and their functional potential across

994 upper, middle and lower stromatolite formations at Cape Recife and Schoenmakerskop.

995

996 Table S1. Summary of genomes binned per collected stromatolite sample

997

998 Table S2. Conserved bacterial species defined by shared ANI greater than 97% in

999 stromatolite formations from Cape Recife and Schoenmakerskop.

1000

1001 Table S3. Summary of characteristics and taxonomic classifications of genomes binned

1002 from shotgun metagenomic data from sampled upper, middle and lower stromatolite

1003 formations from Cape Recife and Schoenmakerskop. Bacterial bins are listed first, with

1004 the two archaeal bins listed at the bottom.

bioRxiv preprint doi: https://doi.org/10.1101/818625; this version posted March 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. bioRxiv preprint doi: https://doi.org/10.1101/818625; this version posted March 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. bioRxiv preprint doi: https://doi.org/10.1101/818625; this version posted March 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. bioRxiv preprint doi: https://doi.org/10.1101/818625; this version posted March 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. bioRxiv preprint doi: https://doi.org/10.1101/818625; this version posted March 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.